Characterization of carbonated serpentine using XPS and TEM

Energy Conversion and Management 45 (2004) 3169–3179www.elsevier.com/locate/enconman

Characterization of carbonated serpentineusing XPS and TEM

Roland K. Schulze, Mary Ann Hill *, Robert D. Field, Pallas A. Papin,Robert J. Hanrahan, Darrin D. Byler

Materials Science and Technology Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA

Received 17 November 2003; accepted 4 February 2004

Available online 18 March 2004

Abstract

With the increasing concentration volume of carbon dioxide in the atmosphere, there has been anincreasing interest in carbon dioxide sequestration. One method is to store the carbon dioxide in mineral

form, reacting solution dissolved CO2 to precipitate carbonates. In order to understand whether or not such

an endeavor is feasible, the carbonation reaction must first be understood. In this study, the surface of

ground serpentine, untreated, heat treated and following a carbonation experiment, has been characterized

using X-ray photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM). The results

indicate that the mechanism for the reaction involves dissolution of the serpentine through the formation of

an amorphous phase and subsequent precipitation of magnesite. The rate limiting step appears to be the

diffusion of Mg out of the amorphous phase.� 2004 Elsevier Ltd. All rights reserved.

Keywords: X-ray photoelectron spectroscopy; Serpentine; Lizardite; Carbonation; CO2 sequestration; Transmission

electron microscopy

1. Introduction

There is increasing interest in studying the disposal of carbon dioxide, one of the primarygreenhouse gases linked to global warming. The successful disposal of carbon dioxide could havea great impact on global warming. One method currently being investigated is chemical fixation ofcarbon dioxide in the form of carbonate minerals [1,2]. The process involves a solution of sodium

* Corresponding author. Fax: +1-505-667-2264.

E-mail address: [email protected] (M.A. Hill).

0196-8904/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.enconman.2004.02.003

mail to: [email protected]

3170 R.K. Schulze et al. / Energy Conversion and Management 45 (2004) 3169–3179

bicarbonate (NaHCO3), sodium chloride (NaCl) and water mixed with serpentine mineral powder(Mg3Si2O5(OH)4). Carbon dioxide is then dissolved into the slurry at an elevated temperature andpressure to enhance the kinetics of the reaction. The result is conversion of the carbon dioxide to athermodynamically stable form, magnesite (MgCO3). Seifritz first proposed this as a naturalmeans of storing CO2 generated from the burning of fossil fuels [3]. With this method, envi-ronmentally benign and thermodynamically stable waste products are created. There is no risk ofreleasing CO2 into the atmosphere at the disposal site as it is chemically bound in a naturalmineral form.

Calcium and magnesium oxides are the only common oxides that readily form carbonates.Magnesium bearing deposits are frequently richer than deposits containing calcium oxide. Thus,magnesium bearing minerals are more attractive for CO2 sequestration. The mineral chosen forthis study is serpentine, with a structure consisting of alternating silicate tetrahedral sheets andbrucite octahedral sheets. The three polymorphs of serpentine are antigorite, lizardite andchrysotile. The structure of antigorite is corrugated; lizardite consists of topological layer;and chrysotile is cylindrical [4]. This mineral, in the purest form, contains 43 wt.% MgO, thestarting material for the carbonation process.

The carbonation of serpentine in natural deposits occurs slowly. To develop an economicallyfeasible carbon sequestration process, the reaction kinetics must be accelerated. This has beenachieved by using elevated temperatures and pressures, adding sodium chloride and sodiumbicarbonate to the aqueous solution, grinding the serpentine to increase the surface area and heattreating the serpentine prior to the carbonation reaction to cause dehydroxylation and therebyincrease the magnesium oxide content by weight in the starting material. Studies indicate that heattreated (dehydroxylated) samples carbonate more readily than untreated samples [5]. A dramaticincrease in carbonation reactivity is associated with samples heat treated at 630 �C, with highertemperatures providing little increase in reactivity [6]. The carbonation process begins whencarbon dioxide is dissolved in water to form carbonic acid (H2CO3). The carbonic acid dissociatesinto Hþ and HCO�

3 . The Hþ ions are then exchanged with Mg2þ cations on the mineral surfaceforming silicic acid (H2O3Si) or silica (SiO2) and water. The free Mg2þ cations then react withbicarbonate ions to form magnesium carbonate (MgCO3). The proposed carbonation reaction isas follows:

Mg3Si2O5ðOHÞ4ðsÞ þ 3CO2ðgÞ ! 3MgCO3ðsÞ þ 2SiO2 þ 2H2OðlÞ

DHreaction ¼ �63:6 kJ=mol CO2 ðexothermicÞ

O�Connor et al. describes aqueous media mineral carbonation in greater detail [7,8].In order for the carbonation process to proceed, MgO needs to be removed from the serpentine.

However, if a mineral dissolves incongruently or nonstoichiometrically, a surface layer can formwith a composition different than that of the bulk. Such a layer may inhibit the diffusion of ionsacross the interface between the mineral surface and the solution. Thus, this layer can be the ratelimiting feature in the carbonation process. Luce et al. noted the incongruent dissolution ofmagnesium silicate [9]. The work of Lin and Clemency [10] supported this research, finding thatmagnesium from the octahedral sheets in antigorite, talc and phlogopite was more likely to bereleased than silicon from the tetrahedral sheet. Tartaj et al. [11] observed a lack of Mg in particle

R.K. Schulze et al. / Energy Conversion and Management 45 (2004) 3169–3179 3171

surface layers during dissolution experiments, resulting in low isoelectric point values in someserpentine samples.

The purpose of this study is to understand the dissolution of the serpentine as a result of thecarbonation process. To accomplish this, it is imperative to characterize the surface of thematerials. The current investigation has utilized X-ray photoelectron spectroscopy (XPS) andtransmission electron microscopy (TEM) to characterize the surface of the carbonated powders.

2. Experimental procedure

Bulk serpentine of the lizardite polymorph was crushed using a jaw crusher set to its smallestopening to produce the finest size fraction of material possible. The crushed material was thenground to a size passing a 200 mesh sieve (157 lm) in a Spex shatterbox. This process was re-peated until sufficient material was produced to complete several experiments. Size fractionationof some of the powder was accomplished using gravity separation. A dispersant of sodiummetahexaphosphate (NaPO3)6 was used during the process. Heat treated samples were placed in aLindberg thermolyne furnace, evacuated to 300 mTorr, backfilled with CO2 gas, and then held at630 �C for 3 h.

The direct solution carbonation experiments used a Haskel 30-152 H two stage, high pressuregas booster to achieve pressures up to 5000 psi. The following materials were placed in the Parrautoclave system: 69.64 g of serpentine, 850 cc of water (serpentine is hydrophilic), 1 M NaCl (aflux), and 0.64 M NaHCO3 (the maximum solubility at room temperature). The vessel was thensealed and purged with CO2. The autoclave was operated at 2300 psi and 155 �C for 2 h. Thesample mixture was stirred at 1490 rpm. Three samples were analyzed using XPS: untreatedserpentine (as-ground), serpentine heat treated at 630 �C in flowing CO2 for 3 h and heat treatedand subsequently solution carbonized material with a 22% conversion to carbonate.

A Physical Electronics 5600ci XPS analyzer was used with Al Ka X-rays (1486.6 eV). Thebinding energies were corrected to the C1s peak at 284.8 eV. Atomic concentrations were cal-culated using the following relative sensitivity factors: C1s (0.314), O1s (0.733), Na1s (1.102),Mg2s (0.274), Si2p (0.368) and Fe2p (2.946). The serpentine powders were mounted on conductivecarbon tape. The sampling depth into the surface of the serpentine particles was approximately50–70 �A . In addition various monolithic standards (MgO, Mg(OH)2, MgCO3) were procured orproduced for spectral comparison purposes.

Specimens for TEM examination were prepared from material sieved to the 0.25–5 lm diameterrange. Selection of this small size fraction was necessary for electron transparency. The powderwas dispersed in isopropyl alcohol using an ultrasonic bath, and samples were floated onto holeycarbon Cu grids. These were analyzed in a Phillips CM30 instrument, operating at 300 KeV,equipped with a PGT thin window EDS detector.

3. Results

A scanning electron micrograph of the starting serpentine powder is shown in Fig. 1. Theparticles are irregular in size and shape with relatively smooth edges and surfaces, resulting from

Fig. 1. Scanning electron micrograph of the serpentine starting powder, 18–40 lm size fraction.


fracture during attrition. XPS survey scans of the three materials, shown in Fig. 2, were quali-tatively the same, although subtle differences in the relative intensities of the Na, Mg and C peaks

0200400600800100012000

0.5

1

1.5

2

2.5

Binding Energy (eV)

Nor

mal

ized

Inte

nsity

-Si2

s

-Si2

p

-Mg

KLL

-Mg2

s -M

g2p

-O K

LL -O1s

-O2s

-Na

KLL

-Na1

s

-Fe

LMM

2

-Fe

LMM

1

-Fe

LMM

-Fe2

p

-C1s

untreated

heattreated

solutiontreated

Fig. 2. XPS survey scans of serpentine in the untreated, heat-treated, and solution-treated conditions.


were observed. A close examination shows enhanced Si relative to Mg at the surface of thesolution treated sample as compared to the other samples. These results immediately suggest thatthe autoclave solution treatment preferentially leaches Mg from the surface, leaving a layer rich inSi.

Representative high resolution XPS results for the Si2p peak and the Mg2s peak for the threeclasses of sample treatment are shown in Fig. 3. The broad peaks suggest that a distribution ofchemical states is present. The salient features of this experiment indicate that while the heattreatment causes a slight reduction in the average Si chemical state (shift to lower binding energy)and a slight oxidation in the average Mg chemical state (shift to higher binding energy), therelative amounts of Si and Mg remain the same as in the untreated sample. The binding energy ofthe Si2p peak for these two samples is approximately 103.0 eV, consistent with silicon in a silicatematrix [12]. Upon CO2 solution treatment, the surface becomes enriched in Si relative to Mg byalmost a factor of two over the untreated material. This silicon rich surface layer may be a passivelayer (diffusion barrier) that results in poor reaction conversion in the bulk of the particle. Inaddition, the Si2p peak shifts to a higher binding energy of 103.5 eV, consistent with the for-mation of silica (SiO2) [12]. The Mg2s binding energy is relatively invariant with chemical state, sochemical changes can be difficult to detect.

Fig. 4 shows a high resolution scan of the O1s XPS peak. The heat treated sample shows alower O1s binding energy (531.7 eV) than the untreated sample (532.0 eV), consistent with theremoval of hydroxyl groups (which appear at higher binding energy in the O1s) at the surface. Thesolution treated sample shows a higher O1s binding energy (532.5 eV) than the untreated sample

8590951001051100

2000

4000

6000

8000

10000

12000

14000

Binding Energy (eV)

c/s

untreated

heattreated

solutiontreated

XPS--Si2p and Mg2sSi2p

Fe3s

Mg2s Mg/Siatomic

1.14

1.15

0.61

Fig. 3. XPS analysis for various treatments of serpentine. The Si2p and Mg2s peaks are indicated. Note relative

intensities of these two peaks. Surface atomic concentration ratios of Mg/Si are also shown for the various treatments.

525530535540

0

1

2

3

4

5

6

7

x 104

Binding Energy ( eV)

c/s

O 1s XPS

untreated

heattreated

solutiontreated

Fig. 4. A high resolution scan of the O1s peak.


(532.0 eV), indicating a possible addition of hydroxyl groups to the silicon rich layer at thesurface. Again, the broad peaks are indicative of a range of chemical states present.

A high resolution scan of the C1s XPS peak shows the presence of a high binding energy featureat 290.5 eV that is undoubtedly a carbonate. Although it was difficult to interpret the C1s peaksunambiguously due to the sample handling and exposure of the sample to the atmosphere afterreaction (different amounts of adventitious carbon from sample handling), the observed chemicalshift of the carbonate feature indicates that this is an ionic metal carbonate, as opposed to anorganic carbonate.

Table 1 shows the atomic concentration results from the XPS scan. The carbon concentrationsare not necessarily representative of the serpentine samples due to preparation artifacts from themounting tape or handling prior to analysis. The Na and Fe concentrations increase followingsolution treatment. The Mg:Si ratio decreases after solution treatment by almost a factor of 2,indicating that the surface is depleted in Mg relative to Si after solution treatment.

Table 1

The atomic concentrations of the surface of serpentine samples using XPS

C O Na Mg Si Fe Mg/Si

Untreated 46.7 37.7 0.4 7.7 6.8 0.8 1.14

Heat treated 24.2 49.8 0.3 13.2 11.5 1.0 1.15

Solution treated 31.0 47.8 1.5 6.8 11.2 1.5 0.61

Fig. 5. TEM micrograph of 2.5–5 lm serpentine particles following carbonation experiment. Particles are mounted on

a holey carbon grid.


TEM analysis of 0.25–5 lm solution treated serpentine powder revealed three predominantfeatures: magnesite particles, lizardite particles and an amorphous phase. The magnesite and li-zardite phases were identified by selected area electron diffraction (SAD). The magnesite particlesdisplayed strong diffraction contrast and were generally highly faceted. The most prominent facetswere on the {1 0 �1 4} planes, a well known growth plane for this structure [13] 1. The lizarditeparticles displayed weaker diffraction contrast and more diffuse reflections in the diffractionpatterns, which tended to deteriorate under the electron beam. These particles were generallysurrounded by the amorphous phase, while the magnesite particles were more likely to havesurfaces free from the amorphous material. TEM micrographs displaying all three of these fea-tures are shown in Figs. 5 and 6. The EDS spectra corresponding to points 1 (magnesite), 2(amorphous phase) and 3 (lizardite) in Fig. 6 are shown in Fig. 7. The Cu signal is an artifact fromthe Cu grid on which the particles were deposited. Mg, Si and O are noted in the amorphous phaseand the lizardite, while the Si peak is absent in the magnesite. A small Fe peak is associated withall of the phases.

4. Discussion

Luce observed three types of kinetic behavior in studying the dissolution kinetics of magnesiumsilicates [9]. The first was rapid surface ion exchange corresponding to a layer thinner than oneunit cell. It has been proposed by Guthrie et al. [14] that the carbonation process is dominated bydissolution and precipitation as opposed to direct carbonation of the magnesium silicate.

1 Indices are for the X-ray smallest cell, ahex ¼ 0:463 nm, chex ¼ 1:502 nm.

Fig. 6. Higher magnification TEM micrograph of area in Fig. 5, showing magnesite particle in the bottom left corner

and lizardite particles surrounded by an amorphous phase in the upper right corner.

0 1 2 3 4 5 6 7

Cou

nts

Energy (KeV)

O

C Cu

MgSi

Fe

Magnesite

Amorphous Phase

Lizardite

Fig. 7. EDS spectra from areas in Fig. 6 for (1) large crystalline magnesite, (2) amorphous phase, and (3) lizardite.



Although Mg2þ may be released through rapid cation exchange with 2Hþ, this effect is assumed tobe minimal since the silicate structure is unsuitable for rapid cation diffusion at low temperatures.The second type of kinetic behavior is slow linear extraction of cations in strong acid solutions. Inthe current study, the pH was not measured in situ during the carbonation process, but it isprobably dominated by bicarbonate (mid pH) and perhaps carbonic acid (low pH) [14]. Thus, thismechanism is unlikely to be operative. The third type of kinetic behavior is slow parabolic thinfilm diffusion controlled cation extraction in mild acid to alkaline solutions. Over longer periods,Luce observed parabolic kinetics at all solution pH values except 1.65 [9]. The parabolic kineticsare consistent with diffusional transport limited processes. The instantaneous rate can be affectedby mineral grinding as the dissolution rates of deformed surfaces and fine particles will be highinitially [15–18]. The dissolution rates then become linear with time, consistent with either surfacelimited reactions or diffusion controlled reactions across a layer of constant thickness:

Rate ¼ dC=dt ¼ kpt

where kp is the reaction rate constant, and t is time [19].If congruent dissolution were to occur, a Mg:Si ratio of 1.5 is expected. The Mg:Si ratio for the

solution treated sample, shown in Table 1, is 0.6. This is indicative of incongruent dissolution thathas been observed for lizardite dissolved under similar conditions by Luce et al. and Guthrie et al.[9,14]. In chemical weathering, kinetic models have found that incongruent dissolution occursinitially with the rapid migration of a specific constituent [20]. In this case, magnesium diffusesrapidly into the aqueous phase leaving behind a layer enriched in silicon. After this initial periodof incongruent dissolution, dissolution is expected to proceed congruently.

A significant qualitative result from the TEM studies is the ubiquitous nature of the Mg peak;specifically, the failure to find any amorphous phase with just Si and O and no Mg. Nonetheless, itis clear that the presence of the amorphous silica layer on the surface of the serpentine particlesaffects the carbonation rate and mechanism. The thickness of the layer ranges from 40 to 100 nm,based on the TEM images. In order for the reaction to proceed, Mg would have to diffuse acrossthe layer of amorphous silica. The transport of cations in silica is so slow that it is usually notmeasurable at temperatures below 1000 �C. Even in polycrystalline quartz, cation transportoccurs principally along defects and grain boundaries rather than through lattice transport. It isapparent that the Mg noted in the amorphous silica layer is trapped and will not diffuse to thesurface to be carbonated in a realistic time frame. Therefore, unless the solution chemistry can bemodified to dissolve silica (e.g. through addition of F� to the solution), then the reaction will notapproach completion for particles of this size in any reasonable time frame.

Two explanations can be suggested for the ubiquitous presence of Mg in the amorphous phase.The first is simply that the reaction does not go to completion, so that the byproduct is a mixedMg–Si oxide (amorphous silica with Mg in solution), however this simple mechanism does notexplain all of the observed microstructure. Therefore, we propose that the initial step in theprocess is the breakdown of the lizardite crystal structure. The magnesite then grows from thisamorphous product, with the rate limiting step being the diffusion of Mg out of the amorphousphase. In this case, the reaction still does not go to completion, as it is limited by the transport ofMg (presumably Mg2þ cations) through amorphous silica. This latter mechanism is then limitedby the formation of a continuous silica layer (with Mg in solution) that acts as a barrier to furtherdiffusion. This would not only explain the absence of pure SiO2, but the frequent observation of


pure magnesite particles with little or no surrounding amorphous phase (i.e. the Mg diffuses out ofthe amorphous phase, forming magnesite on the edges, rather than the SiO2 being pushed outof the growing magnesite). De la Caillerie et al. detected an amorphous silicate during contact ofsolvated Mg2þ cations with silica and moderately basic pH [21]. A Si–O–Mg bond was found withnuclear magnetic resonance (NMR), showing that the Mg was fixed through the formation of ahigh surface area, amorphous phase described as a magnesium silicate gel layer on a silica surface.In De la Caillerie�s study, there was no evidence of a dissolution–precipitation kinetic effect wherethe absorption of the magnesium disturbed the chemical environment to such an extent that theSi–O bonds were weakened. These observations are consistent with the proposed mechanism aswell as the observations in XPS where, upon solution treatment, the Si chemical form changesfrom one that is silicate-like to one that is silica-like.

5. Conclusions

The carbonation of ground serpentine was studied to determine the feasibility of convertingcarbon dioxide to a thermodynamically stable form, magnesite (MgCO3). Serpentine before andafter treatment in a CO2 reaction vessel was characterized using X-ray photoelectron spectroscopyand transmission electron spectroscopy to provide a better understanding of the surface reactionsoccurring during the carbonation process. The XPS and TEM data show that a Mg containingsilica layer is formed around the serpentine following carbonation. Diffusion of magnesium fromthis amorphous phase is believed to be the rate limiting step in the carbonation reaction. In orderfor this process to be used for carbon dioxide disposal, the solution chemistry must be modified tofacilitate dissolution of this amorphous silica layer.

Acknowledgements

We gratefully acknowledge the support of Los Alamos National Laboratory through thelaboratory directed research and development program. We thank Hans Ziock and GeorgeGuthrie for many useful discussions.

References

[1] Lackner KS, Butt DP, Wendt CH. Progress on binding CO2 in mineral substrates. Energy Convers Manage

1997;38:S259–64.

[2] Lackner KS, Wendt CH, Butt DP, Joyce EL, Sharp DH. Carbon dioxide disposal in carbonate minerals. Energy

1995;20(11):1153–70.

[3] Seifritz W. CO2 disposal by means of silicates. Nature 1990;345:486.

[4] Wicks FJ, O�Hanley DS. In: Reviews in mineralogy serpentine minerals: structures and petrology, vol 19.

Washington, DC: Mineral Society of America; 1988.

[5] McKelvy M, Sharma R, Chizmeshya A, Carpenter R, Streib K. Magnesium hydroxide dehydroxylation: In situ

nanoscale observations of lamellar nucleation and growth. Chem Mater 2001;13:921–6.


[6] O�Connor WK, Dahlin DC, Nilsen DN, Rush GE, Walters RP, Turner PC. Carbon dioxide sequestration by direct

mineral carbonation: Results from recent studies. In: Proceedings of the First National Conference on Carbon

Sequestration, Washington DC, NETL, 2001.

[7] O�Connor WK, Dahlin DC, Turner PC, Walters RP. Carbon dioxide sequestration by ex-situ mineral carbonation.

Technology 1999;7S:115–23.

[8] O�Connor WK, Dahlin DC, Nilsen DN, Rush GE, Walters RP, Turner PC. CO2 storage in solid form: A study of

direct mineral carbonation. In: Proceedings of the 5th International Conference on Greenhouse Gas Technologies,

Cairns, Australia, 2000. p. 7.

[9] Luce RW, Bartlett RW, Parks GA. Dissolution kinetics of magnesium silicates. Geochim Cosmochim Acta

1972;36:35–50.

[10] Lin F, Clemency CV. The dissolution kinetics of brucite, antigorite, talc, and phlogopite at room temperature and

pressure. Am Mineral 1981;66:801–6.

[11] Tartaj P, Cerpa A, Garica-Gonzalez MT, Serna CJ. Surface instability of serpentine in aqueous suspensions. J

Colloid Interface Sci 2000;231:176–81.

[12] Moulder JF, Stickle WF, Sobol PE, Bomben KD. Handbook of X-ray photoelectron spectroscopy. Eden Prairie,

Minnesota: Perkin–Elmer; 1992. and references therein.

[13] Teng H, Dove PM, De Yoreo JJ. Kinetics of calcite growth: Surface processes and relationships to macroscopic

rate laws. Geochim Cosmochim Acta 2000;64:2255–66.

[14] Guthrie GD, Carey JW, Bergfeld D, Byler D, Chipera S, Ziock HJ, et al. Geochemical aspects of the carbonation

of magnesium silicates in an aqueous medium. In: Proceedings of the First National Conference on Carbon

Sequestration, Washington DC, NETL, 2001.

[15] Berner RA, Holdren Jr GR. Mechanism of feldspar weathering-II. Observations of feldspars from soils. Geochim

Cosmochim Acta 1979;43:1173–86.

[16] Berner RA, Sjoberg EL, Velbel MA, Krom MD. Dissolution of pyroxenes and amphiboles during weathering.

Science 1980;207:1205–6.

[17] Holdren Jr GR, Berner RA. Mechanism of feldspar weathering -I. Experimental studies. Geochim Cosmochim

Acta 1979;36:1067–70.

[18] Schott J, Berner RA, Sjoberg EL. Mechanism of pyroxene and amphibole weathering-I. Experimental studies of

iron-free minerals. Geochim Cosmochim Acta 1981;43:1161–71.

[19] Stumm W, Morgan JJ. Aquatic chemistry. New York: John Wiley and Sons; 1996.

[20] Schnoor JL. The kinetics of chemical weathering: a comparison of laboratory and field weathering rates. In: Stumm

W, editor. Aquatic chemical kinetics. New York: John Wiley and Sons; 1990. p. 476.

[21] De la Caillerie JBD, Kermarec M, Clause O. Si NMR Observation of an amorphous magnesium silicate formed

during impregnation of silica with Mg(II) in aqueous solution. J Phys Chem 1995;99:17273–81.

Documents

Characterization of carbonated serpentine using XPS and TEM